path: root/resources/tools/presentation/generator_plots.py

# Copyright (c) 2018 Cisco and/or its affiliates.
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at:
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

"""Algorithms to generate plots.
"""


import logging
import pandas as pd
import plotly.offline as ploff
import plotly.graph_objs as plgo

from plotly.exceptions import PlotlyError
from collections import OrderedDict
from copy import deepcopy

from utils import mean


COLORS = ["SkyBlue", "Olive", "Purple", "Coral", "Indigo", "Pink",
          "Chocolate", "Brown", "Magenta", "Cyan", "Orange", "Black",
          "Violet", "Blue", "Yellow", "BurlyWood", "CadetBlue", "Crimson",
          "DarkBlue", "DarkCyan", "DarkGreen", "Green", "GoldenRod",
          "LightGreen", "LightSeaGreen", "LightSkyBlue", "Maroon",
          "MediumSeaGreen", "SeaGreen", "LightSlateGrey"]


def generate_plots(spec, data):
    """Generate all plots specified in the specification file.

    :param spec: Specification read from the specification file.
    :param data: Data to process.
    :type spec: Specification
    :type data: InputData
    """

    logging.info("Generating the plots ...")
    for index, plot in enumerate(spec.plots):
        try:
            logging.info("  Plot nr {0}: {1}".format(index + 1,
                                                     plot.get("title", "")))
            plot["limits"] = spec.configuration["limits"]
            eval(plot["algorithm"])(plot, data)
            logging.info("  Done.")
        except NameError as err:
            logging.error("Probably algorithm '{alg}' is not defined: {err}".
                          format(alg=plot["algorithm"], err=repr(err)))
    logging.info("Done.")


def plot_performance_box(plot, input_data):
    """Generate the plot(s) with algorithm: plot_performance_box
    specified in the specification file.

    :param plot: Plot to generate.
    :param input_data: Data to process.
    :type plot: pandas.Series
    :type input_data: InputData
    """

    # Transform the data
    plot_title = plot.get("title", "")
    logging.info("    Creating the data set for the {0} '{1}'.".
                 format(plot.get("type", ""), plot_title))
    data = input_data.filter_data(plot)
    if data is None:
        logging.error("No data.")
        return

    # Prepare the data for the plot
    y_vals = dict()
    y_tags = dict()
    for job in data:
        for build in job:
            for test in build:
                if y_vals.get(test["parent"], None) is None:
                    y_vals[test["parent"]] = list()
                    y_tags[test["parent"]] = test.get("tags", None)
                try:
                    if test["type"] in ("NDRPDR", ):
                        if "-pdr" in plot_title.lower():
                            y_vals[test["parent"]].\
                                append(test["throughput"]["PDR"]["LOWER"])
                        elif "-ndr" in plot_title.lower():
                            y_vals[test["parent"]]. \
                                append(test["throughput"]["NDR"]["LOWER"])
                        else:
                            continue
                    else:
                        continue
                except (KeyError, TypeError):
                    y_vals[test["parent"]].append(None)

    # Sort the tests
    order = plot.get("sort", None)
    if order and y_tags:
        y_sorted = OrderedDict()
        y_tags_l = {s: [t.lower() for t in ts] for s, ts in y_tags.items()}
        for tag in order:
            logging.info(tag)
            for suite, tags in y_tags_l.items():
                if "not " in tag:
                    tag = tag.split(" ")[-1]
                    if tag.lower() in tags:
                        continue
                else:
                    if tag.lower() not in tags:
                        continue
                try:
                    y_sorted[suite] = y_vals.pop(suite)
                    y_tags_l.pop(suite)
                    logging.info(suite)
                except KeyError as err:
                    logging.error("Not found: {0}".format(err))
                finally:
                    break
    else:
        y_sorted = y_vals

    # Add None to the lists with missing data
    max_len = 0
    for val in y_sorted.values():
        if len(val) > max_len:
            max_len = len(val)
    for key, val in y_sorted.items():
        if len(val) < max_len:
            val.extend([None for _ in range(max_len - len(val))])

    # Add plot traces
    traces = list()
    df = pd.DataFrame(y_sorted)
    df.head()
    y_max = list()
    for i, col in enumerate(df.columns):
        name = "{0}. {1}".format(i + 1, col.lower().replace('-ndrpdrdisc', '').
                                 replace('-ndrpdr', ''))
        logging.info(name)
        traces.append(plgo.Box(x=[str(i + 1) + '.'] * len(df[col]),
                               y=[y / 1000000 if y else None for y in df[col]],
                               name=name,
                               **plot["traces"]))
        try:
            val_max = max(df[col])
        except ValueError as err:
            logging.error(err)
            continue
        if val_max:
            y_max.append(int(val_max / 1000000) + 1)

    try:
        # Create plot
        layout = deepcopy(plot["layout"])
        if layout.get("title", None):
            layout["title"] = "<b>Packet Throughput:</b> {0}". \
                format(layout["title"])
        if y_max:
            layout["yaxis"]["range"] = [0, max(y_max)]
        plpl = plgo.Figure(data=traces, layout=layout)

        # Export Plot
        logging.info("    Writing file '{0}{1}'.".
                     format(plot["output-file"], plot["output-file-type"]))
        ploff.plot(plpl, show_link=False, auto_open=False,
                   filename='{0}{1}'.format(plot["output-file"],
                                            plot["output-file-type"]))
    except PlotlyError as err:
        logging.error("   Finished with error: {}".
                      format(str(err).replace("\n", " ")))
        return


def plot_latency_error_bars(plot, input_data):
    """Generate the plot(s) with algorithm: plot_latency_error_bars
    specified in the specification file.

    :param plot: Plot to generate.
    :param input_data: Data to process.
    :type plot: pandas.Series
    :type input_data: InputData
    """

    # Transform the data
    plot_title = plot.get("title", "")
    logging.info("    Creating the data set for the {0} '{1}'.".
                 format(plot.get("type", ""), plot_title))
    data = input_data.filter_data(plot)
    if data is None:
        logging.error("No data.")
        return

    # Prepare the data for the plot
    y_tmp_vals = dict()
    y_tags = dict()
    for job in data:
        for build in job:
            for test in build:
                if y_tmp_vals.get(test["parent"], None) is None:
                    y_tmp_vals[test["parent"]] = [
                        list(),  # direction1, min
                        list(),  # direction1, avg
                        list(),  # direction1, max
                        list(),  # direction2, min
                        list(),  # direction2, avg
                        list()   # direction2, max
                    ]
                    y_tags[test["parent"]] = test.get("tags", None)
                try:
                    if test["type"] in ("NDRPDR", ):
                        if "-pdr" in plot_title.lower():
                            ttype = "PDR"
                        elif "-ndr" in plot_title.lower():
                            ttype = "NDR"
                        else:
                            continue
                        y_tmp_vals[test["parent"]][0].append(
                            test["latency"][ttype]["direction1"]["min"])
                        y_tmp_vals[test["parent"]][1].append(
                            test["latency"][ttype]["direction1"]["avg"])
                        y_tmp_vals[test["parent"]][2].append(
                            test["latency"][ttype]["direction1"]["max"])
                        y_tmp_vals[test["parent"]][3].append(
                            test["latency"][ttype]["direction2"]["min"])
                        y_tmp_vals[test["parent"]][4].append(
                            test["latency"][ttype]["direction2"]["avg"])
                        y_tmp_vals[test["parent"]][5].append(
                            test["latency"][ttype]["direction2"]["max"])
                    else:
                        continue
                except (KeyError, TypeError):
                    pass

    # Sort the tests
    order = plot.get("sort", None)
    if order and y_tags:
        y_sorted = OrderedDict()
        y_tags_l = {s: [t.lower() for t in ts] for s, ts in y_tags.items()}
        for tag in order:
            for suite, tags in y_tags_l.items():
                if tag.lower() in tags:
                    try:
                        y_sorted[suite] = y_tmp_vals.pop(suite)
                        y_tags_l.pop(suite)
                    except KeyError as err:
                        logging.error("Not found: {0}".format(err))
                    finally:
                        break
    else:
        y_sorted = y_tmp_vals

    x_vals = list()
    y_vals = list()
    y_mins = list()
    y_maxs = list()
    for key, val in y_sorted.items():
        key = "-".join(key.split("-")[1:-1])
        x_vals.append(key)  # dir 1
        y_vals.append(mean(val[1]) if val[1] else None)
        y_mins.append(mean(val[0]) if val[0] else None)
        y_maxs.append(mean(val[2]) if val[2] else None)
        x_vals.append(key)  # dir 2
        y_vals.append(mean(val[4]) if val[4] else None)
        y_mins.append(mean(val[3]) if val[3] else None)
        y_maxs.append(mean(val[5]) if val[5] else None)

    traces = list()
    annotations = list()

    for idx in range(len(x_vals)):
        if not bool(int(idx % 2)):
            direction = "West - East"
        else:
            direction = "East - West"
        hovertext = ("Test: {test}<br>"
                     "Direction: {dir}<br>".format(test=x_vals[idx],
                                                   dir=direction))
        if isinstance(y_maxs[idx], float):
            hovertext += "Max: {max:.2f}uSec<br>".format(max=y_maxs[idx])
        if isinstance(y_vals[idx], float):
            hovertext += "Avg: {avg:.2f}uSec<br>".format(avg=y_vals[idx])
        if isinstance(y_mins[idx], float):
            hovertext += "Min: {min:.2f}uSec".format(min=y_mins[idx])

        if isinstance(y_maxs[idx], float) and isinstance(y_vals[idx], float):
            array = [y_maxs[idx] - y_vals[idx], ]
        else:
            array = [None, ]
        if isinstance(y_mins[idx], float) and isinstance(y_vals[idx], float):
            arrayminus = [y_vals[idx] - y_mins[idx], ]
        else:
            arrayminus = [None, ]
        traces.append(plgo.Scatter(
            x=[idx, ],
            y=[y_vals[idx], ],
            name=x_vals[idx],
            legendgroup=x_vals[idx],
            showlegend=bool(int(idx % 2)),
            mode="markers",
            error_y=dict(
                type='data',
                symmetric=False,
                array=array,
                arrayminus=arrayminus,
                color=COLORS[int(idx / 2)]
            ),
            marker=dict(
                size=10,
                color=COLORS[int(idx / 2)],
            ),
            text=hovertext,
            hoverinfo="text",
        ))
        annotations.append(dict(
            x=idx,
            y=0,
            xref="x",
            yref="y",
            xanchor="center",
            yanchor="top",
            text="E-W" if bool(int(idx % 2)) else "W-E",
            font=dict(
                size=16,
            ),
            align="center",
            showarrow=False
        ))

    try:
        # Create plot
        logging.info("    Writing file '{0}{1}'.".
                     format(plot["output-file"], plot["output-file-type"]))
        layout = deepcopy(plot["layout"])
        if layout.get("title", None):
            layout["title"] = "<b>Packet Latency:</b> {0}".\
                format(layout["title"])
        layout["annotations"] = annotations
        plpl = plgo.Figure(data=traces, layout=layout)

        # Export Plot
        ploff.plot(plpl,
                   show_link=False, auto_open=False,
                   filename='{0}{1}'.format(plot["output-file"],
                                            plot["output-file-type"]))
    except PlotlyError as err:
        logging.error("   Finished with error: {}".
                      format(str(err).replace("\n", " ")))
        return


def plot_throughput_speedup_analysis(plot, input_data):
    """Generate the plot(s) with algorithm:
    plot_throughput_speedup_analysis
    specified in the specification file.

    :param plot: Plot to generate.
    :param input_data: Data to process.
    :type plot: pandas.Series
    :type input_data: InputData
    """

    # Transform the data
    plot_title = plot.get("title", "")
    logging.info("    Creating the data set for the {0} '{1}'.".
                 format(plot.get("type", ""), plot_title))
    data = input_data.filter_data(plot)
    if data is None:
        logging.error("No data.")
        return

    y_vals = dict()
    y_tags = dict()
    for job in data:
        for build in job:
            for test in build:
                if y_vals.get(test["parent"], None) is None:
                    y_vals[test["parent"]] = {"1": list(),
                                              "2": list(),
                                              "4": list()}
                    y_tags[test["parent"]] = test.get("tags", None)
                try:
                    if test["type"] in ("NDRPDR",):
                        if "-pdr" in plot_title.lower():
                            ttype = "PDR"
                        elif "-ndr" in plot_title.lower():
                            ttype = "NDR"
                        else:
                            continue
                        if "1C" in test["tags"]:
                            y_vals[test["parent"]]["1"]. \
                                append(test["throughput"][ttype]["LOWER"])
                        elif "2C" in test["tags"]:
                            y_vals[test["parent"]]["2"]. \
                                append(test["throughput"][ttype]["LOWER"])
                        elif "4C" in test["tags"]:
                            y_vals[test["parent"]]["4"]. \
                                append(test["throughput"][ttype]["LOWER"])
                except (KeyError, TypeError):
                    pass

    if not y_vals:
        logging.warning("No data for the plot '{}'".
                        format(plot.get("title", "")))
        return

    y_1c_max = dict()
    for test_name, test_vals in y_vals.items():
        for key, test_val in test_vals.items():
            if test_val:
                y_vals[test_name][key] = sum(test_val) / len(test_val)
                if key == "1":
                    y_1c_max[test_name] = max(test_val) / 1000000.0

    vals = dict()
    y_max = list()
    nic_limit = 0
    lnk_limit = 0
    pci_limit = plot["limits"]["pci"]["pci-g3-x8"]
    for test_name, test_vals in y_vals.items():
        if test_vals["1"]:
            name = "-".join(test_name.split('-')[1:-1])

            vals[name] = dict()
            y_val_1 = test_vals["1"] / 1000000.0
            y_val_2 = test_vals["2"] / 1000000.0 if test_vals["2"] else None
            y_val_4 = test_vals["4"] / 1000000.0 if test_vals["4"] else None

            vals[name]["val"] = [y_val_1, y_val_2, y_val_4]
            vals[name]["rel"] = [1.0, None, None]
            vals[name]["ideal"] = [y_1c_max[test_name],
                                   y_1c_max[test_name] * 2,
                                   y_1c_max[test_name] * 4]
            vals[name]["diff"] = \
                [(y_val_1 - y_1c_max[test_name]) * 100 / y_val_1, None, None]

            try:
                val_max = max(max(vals[name]["val"], vals[name]["ideal"]))
            except ValueError as err:
                logging.error(err)
                continue
            if val_max:
                y_max.append(int((val_max / 10) + 1) * 10)

            if y_val_2:
                vals[name]["rel"][1] = round(y_val_2 / y_val_1, 2)
                vals[name]["diff"][1] = \
                    (y_val_2 - vals[name]["ideal"][1]) * 100 / y_val_2
            if y_val_4:
                vals[name]["rel"][2] = round(y_val_4 / y_val_1, 2)
                vals[name]["diff"][2] = \
                    (y_val_4 - vals[name]["ideal"][2]) * 100 / y_val_4

        # Limits:
        if "x520" in test_name:
            limit = plot["limits"]["nic"]["x520"]
        elif "x710" in test_name:
            limit = plot["limits"]["nic"]["x710"]
        elif "xxv710" in test_name:
            limit = plot["limits"]["nic"]["xxv710"]
        elif "xl710" in test_name:
            limit = plot["limits"]["nic"]["xl710"]
        else:
            limit = 0
        if limit > nic_limit:
            nic_limit = limit

        mul = 2 if "ge2p" in test_name else 1
        if "10ge" in test_name:
            limit = plot["limits"]["link"]["10ge"] * mul
        elif "25ge" in test_name:
            limit = plot["limits"]["link"]["25ge"] * mul
        elif "40ge" in test_name:
            limit = plot["limits"]["link"]["40ge"] * mul
        elif "100ge" in test_name:
            limit = plot["limits"]["link"]["100ge"] * mul
        else:
            limit = 0
        if limit > lnk_limit:
            lnk_limit = limit

    # Sort the tests
    order = plot.get("sort", None)
    if order and y_tags:
        y_sorted = OrderedDict()
        y_tags_l = {s: [t.lower() for t in ts] for s, ts in y_tags.items()}
        for tag in order:
            for test, tags in y_tags_l.items():
                if tag.lower() in tags:
                    name = "-".join(test.split('-')[1:-1])
                    try:
                        y_sorted[name] = vals.pop(name)
                        y_tags_l.pop(test)
                    except KeyError as err:
                        logging.error("Not found: {0}".format(err))
                    finally:
                        break
    else:
        y_sorted = vals

    traces = list()
    annotations = list()
    x_vals = [1, 2, 4]

    # Limits:
    try:
        threshold = 1.1 * max(y_max)  # 10%
    except ValueError as err:
        logging.error(err)
        return
    nic_limit /= 1000000.0
    if nic_limit < threshold:
        traces.append(plgo.Scatter(
            x=x_vals,
            y=[nic_limit, ] * len(x_vals),
            name="NIC: {0:.2f}Mpps".format(nic_limit),
            showlegend=False,
            mode="lines",
            line=dict(
                dash="dot",
                color=COLORS[-1],
                width=1),
            hoverinfo="none"
        ))
        annotations.append(dict(
            x=1,
            y=nic_limit,
            xref="x",
            yref="y",
            xanchor="left",
            yanchor="bottom",
            text="NIC: {0:.2f}Mpps".format(nic_limit),
            font=dict(
                size=14,
                color=COLORS[-1],
            ),
            align="left",
            showarrow=False
        ))
        y_max.append(int((nic_limit / 10) + 1) * 10)

    lnk_limit /= 1000000.0
    if lnk_limit < threshold:
        traces.append(plgo.Scatter(
            x=x_vals,
            y=[lnk_limit, ] * len(x_vals),
            name="Link: {0:.2f}Mpps".format(lnk_limit),
            showlegend=False,
            mode="lines",
            line=dict(
                dash="dot",
                color=COLORS[-2],
                width=1),
            hoverinfo="none"
        ))
        annotations.append(dict(
            x=1,
            y=lnk_limit,
            xref="x",
            yref="y",
            xanchor="left",
            yanchor="bottom",
            text="Link: {0:.2f}Mpps".format(lnk_limit),
            font=dict(
                size=14,
                color=COLORS[-2],
            ),
            align="left",
            showarrow=False
        ))
        y_max.append(int((lnk_limit / 10) + 1) * 10)

    pci_limit /= 1000000.0
    if pci_limit < threshold:
        traces.append(plgo.Scatter(
            x=x_vals,
            y=[pci_limit, ] * len(x_vals),
            name="PCIe: {0:.2f}Mpps".format(pci_limit),
            showlegend=False,
            mode="lines",
            line=dict(
                dash="dot",
                color=COLORS[-3],
                width=1),
            hoverinfo="none"
        ))
        annotations.append(dict(
            x=1,
            y=pci_limit,
            xref="x",
            yref="y",
            xanchor="left",
            yanchor="bottom",
            text="PCIe: {0:.2f}Mpps".format(pci_limit),
            font=dict(
                size=14,
                color=COLORS[-3],
            ),
            align="left",
            showarrow=False
        ))
        y_max.append(int((pci_limit / 10) + 1) * 10)

    # Perfect and measured:
    cidx = 0
    for name, val in y_sorted.iteritems():
        hovertext = list()
        for idx in range(len(val["val"])):
            htext = ""
            if isinstance(val["val"][idx], float):
                htext += "value: {0:.2f}Mpps<br>".format(val["val"][idx])
            if isinstance(val["diff"][idx], float):
                htext += "diff: {0:.0f}%<br>".format(round(val["diff"][idx]))
            if isinstance(val["rel"][idx], float):
                htext += "speedup: {0:.2f}".format(val["rel"][idx])
            hovertext.append(htext)
        traces.append(plgo.Scatter(x=x_vals,
                                   y=val["val"],
                                   name=name,
                                   legendgroup=name,
                                   mode="lines+markers",
                                   line=dict(
                                       color=COLORS[cidx],
                                       width=2),
                                   marker=dict(
                                       symbol="circle",
                                       size=10
                                   ),
                                   text=hovertext,
                                   hoverinfo="text+name"
                                   ))
        traces.append(plgo.Scatter(x=x_vals,
                                   y=val["ideal"],
                                   name="{0} perfect".format(name),
                                   legendgroup=name,
                                   showlegend=False,
                                   mode="lines",
                                   line=dict(
                                       color=COLORS[cidx],
                                       width=2,
                                       dash="dash"),
                                   text=["perfect: {0:.2f}Mpps".format(y)
                                         for y in val["ideal"]],
                                   hoverinfo="text"
                                   ))
        cidx += 1

    try:
        # Create plot
        logging.info("    Writing file '{0}{1}'.".
                     format(plot["output-file"], plot["output-file-type"]))
        layout = deepcopy(plot["layout"])
        if layout.get("title", None):
            layout["title"] = "<b>Speedup Multi-core:</b> {0}". \
                format(layout["title"])
        layout["annotations"].extend(annotations)
        plpl = plgo.Figure(data=traces, layout=layout)

        # Export Plot
        ploff.plot(plpl,
                   show_link=False, auto_open=False,
                   filename='{0}{1}'.format(plot["output-file"],
                                            plot["output-file-type"]))
    except PlotlyError as err:
        logging.error("   Finished with error: {}".
                      format(str(err).replace("\n", " ")))
        return


def plot_http_server_performance_box(plot, input_data):
    """Generate the plot(s) with algorithm: plot_http_server_performance_box
    specified in the specification file.

    :param plot: Plot to generate.
    :param input_data: Data to process.
    :type plot: pandas.Series
    :type input_data: InputData
    """

    # Transform the data
    logging.info("    Creating the data set for the {0} '{1}'.".
                 format(plot.get("type", ""), plot.get("title", "")))
    data = input_data.filter_data(plot)
    if data is None:
        logging.error("No data.")
        return

    # Prepare the data for the plot
    y_vals = dict()
    for job in data:
        for build in job:
            for test in build:
                if y_vals.get(test["name"], None) is None:
                    y_vals[test["name"]] = list()
                try:
                    y_vals[test["name"]].append(test["result"])
                except (KeyError, TypeError):
                    y_vals[test["name"]].append(None)

    # Add None to the lists with missing data
    max_len = 0
    for val in y_vals.values():
        if len(val) > max_len:
            max_len = len(val)
    for key, val in y_vals.items():
        if len(val) < max_len:
            val.extend([None for _ in range(max_len - len(val))])

    # Add plot traces
    traces = list()
    df = pd.DataFrame(y_vals)
    df.head()
    for i, col in enumerate(df.columns):
        name = "{0}. {1}".format(i + 1, col.lower().replace('-cps', '').
                                 replace('-rps', ''))
        traces.append(plgo.Box(x=[str(i + 1) + '.'] * len(df[col]),
                               y=df[col],
                               name=name,
                               **plot["traces"]))
    try:
        # Create plot
        plpl = plgo.Figure(data=traces, layout=plot["layout"])

        # Export Plot
        logging.info("    Writing file '{0}{1}'.".
                     format(plot["output-file"], plot["output-file-type"]))
        ploff.plot(plpl, show_link=False, auto_open=False,
                   filename='{0}{1}'.format(plot["output-file"],
                                            plot["output-file-type"]))
    except PlotlyError as err:
        logging.error("   Finished with error: {}".
                      format(str(err).replace("\n", " ")))
        return